SESSION TITLE: Decision-Making in Lung Cancer
SESSION TYPE: Original Investigation Slide
PRESENTED ON: Wednesday, October 30, 2013 at 07:30 AM - 09:00 AM
PURPOSE: Screening with LDCT reduces lung cancer mortality in the population defined by the National Lung Screening Trial (NLST). Screening of non-NLST populations is recommended by several guidelines, though benefits are unproven. Our group sought to compare physician estimates of the probability of lung cancer with published prediction models, and to determine whether presentation of the risk model results influenced physician decisions in considering lung cancer screening.
METHODS: We surveyed physicians visiting the Yale Lung SCAN CHEST 2012 Center of Excellence exhibit using case vignettes (two that met NLST criteria and three that did not); participants were asked whether they would perform screening and to estimate lung cancer risk over 5 and 10 years. Physicians then viewed a tablet application generating risk predictions from five published lung cancer predictive models, and their decisions on whether to screen for lung cancer were again solicited.
RESULTS: 102 physicians participated. For the two cases meeting the NLST criteria, 69% and 74% of physicians recommended screening. Physician estimates of 10-year lung cancer risk were 23% and 15%, compared to 6% and 11% predicted by the Bach model, respectively. After viewing the results from the model, 75% and 91% of physicians recommended screening, respectively. For the three non-NLST cases screening was recommended by 56%, 43%, and 84% of physicians; estimates of 5-year lung cancer risk were 15%, 11%, and 14%, compared to 0.1%, 8.5%, and 6.5% predicted by the Liverpool model, respectively. After viewing the model predictions, 31%, 57%, and 72% of physicians recommended screening, respectively.
CONCLUSIONS: Compared to validated prediction models, physicians overestimated the risk of developing lung cancer. Physician decisions relating to screening appeared to be influenced by results of the prediction models.
CLINICAL IMPLICATIONS: Clinical prediction models have the potential to inform physicians regarding individual risk of lung cancer and may influence decision-making regarding lung cancer screening. As the models become more user friendly, they may play an important role in comprehensive lung cancer screening programs.
DISCLOSURE: Frank Detterbeck: Consultant fee, speaker bureau, advisory committee, etc.: Lilly - memeber of international staging committe and lectures about staging. , Consultant fee, speaker bureau, advisory committee, etc.: Oncimmune - avisory panel to review data, Consultant fee, speaker bureau, advisory committee, etc.: Pfizer - external grant administration board 2012, Grant monies (from industry related sources): Medela - particpated in a multicenter study of hte value of Medela thoracic drainage collection devices. Particpated as an advisory board of experts in 2009. , Consultant fee, speaker bureau, advisory committee, etc.: Covidien- Advisory board member 2009 , Grant monies (from industry related sources): Deepbreeze - PI of multicenter study to assess how well VRI can predict regional lung function and posoperative lung fucntion in patiens who are undergoing resection for lung cancer. The following authors have nothing to disclose: Amanda Reid, Gaetane Michaud, Paul Guillod, Lynn Tanoue
No Product/Research Disclosure Information